This open-source project concerns the public german television. They host on-demand news, whereby the online available video contains blocked frames. Connected to a Raspberry Pi, herewith the blocekd fraction is measured, documented and posted online. For detailed informations please check out the website, twitterbot or mastodon account, as this README.md focuses on the technical aspect.
To run the software writen in R, you'll aditionally have to install some Linux libraries:
sudo apt install r-base ffmpeg libcurl4-openssl-dev libssl-dev libssh2-1-dev imagemagick libmagick++-dev git
Next install the depending R packages:
install.packages(c("jpeg", "rvest", "ggplot2", "tibble", "lubridate", "stringr", "magick")).
Further, for communicating the results:
install.packages("twitteR", "telegram.bot")
and the Mastodon package not on CRAN:
devtools::install_github('ThomasChln/mastodon') # install.packages("devtools")
Simply clone this repo with git clone git@github.com:georoen/standbildNews.git.
Now give it a spin by e.g. processing the latest ZDF 19Uhr news:
Rscript --vanilla bin/MAIN.R h19 2
The syntax is Rscript --vanilla for sourcing the R script without saving the enviroment, bin/heuteZensiert.R calling the main, h19 defining the news broadcast you want to stream and 2 for selecting the broadcast of two days ago.
Along h19 for Heute 19Uhr, also the other show of ZDF hjo for Heute Journal, as well as the two shows of ARD, t20 for Tagesschau and tth Tagesthemen are implemented.
The number 2 is an optional date parameter, selecting the show of two days ago.
Last but not least, you can add the following processes to crontab:
0 20 * * * cd ~/standbildNews/ && nohup Rscript --vanilla bin/MAIN.R h19 &
0 21 * * * cd ~/standbildNews/ && nohup Rscript --vanilla bin/MAIN.R t20 &
0 22 * * * cd ~/standbildNews/ && nohup Rscript --vanilla bin/MAIN.R tth &
0 23 * * * cd ~/standbildNews/ && nohup Rscript --vanilla bin/MAIN.R hjo &
Whereby technically, you'll have to overwrite the social credentials stored in extra/, as well as need write accees to Logfile.csv. Please don't hesitate to contact us.
There are several ways how to contribute to this project:
- Are the tweets correct? If you watch the above news on a regular basis and have a social account, please fave/interact with, if the blocked frames were detected correct. This helps to access the classifieres accuracy.
- Add context to the blocked frames. It is not only interesting to know how much frames are missing, put also what is missing. Thus when you post a short reply about the blocked content, that would be very cool.
- Run a node. If you have a Raspberry Pi or any other server you can help processing the shows every day. Please do not hestitate to contact us, so we can share the credentials with you.